View Full Version : Evox rifa MMK feedback caps

mega lord
02-05-07, 12:36 AM

Over the last three weeks i have replaced the feedback caps in my nap140 and nac72 321 boards with 22uf MMK's.

In the 140 i have used sections of rubber eraser cut into oblongs and hotmelt glued to the board to provide a mount for the caps to be glued on (on their sides. This raises them so they are clear of surrounding components and flying leads of ptfe silver wire are used in the shortest length possible. This mounting method seems to work and the properties of rubber as a material with regards to vibration should be acceptable. Do you agree?

With the nac 72 i scratched my head a fair bit when trying to work out the best place to mount the mmk's. I had to take into account the fact that 10uf smr's will be used as signal caps at some point so , i made sure enough room was available for them in the future. The result was the mmks hotglued on top of the volume pot. Is this acceptable do you think? Will i be getting any horrible crosstalk or nasty effects from the pot? (feedback resistor was changed to 2K welwyn rc55y).

As regards the sound - at once it was evident there was more detail coming through and much greater clarity, all with the PRaT intact. The change in bass is awesome - at first it seemed like there was less of it coming through but what has actually happened is that a degree of bloatedness has been stripped away which enables every note to be heard in its own right. This was backed up by listening to some Jaco Pastorius - some of his fast bass runs did'nt quite make sense before, but now you can clearly hear every note stop and start, harmonics are gorgeous and i think i know what bass guitar he used.
Treble is equally as good.

Will i be going too far if i replace all the signal caps with smr's? I dont want to ruin the sound and create more work by having to pull them out and use somethin else .

thanks stu

02-05-07, 04:54 AM
...the properties of rubber as a material with regards to vibration should be acceptable. Do you agree?

It's really impossible to say! There is a lot of rubbish talked about this kind of thing. To answer the question properly you would need to know:-
Are the capacitors sensitive to vibration?
Which frequencies are important?
What are the dynamic properties of the rest of the PCB, the case, etc?
How much vibration is around and what frequencies?
What is the mass of the capacitor?
What are the dynamic properties of the rubber and what shape and size is it?
...and that's just a starting point.
Then you could design the mounting to make sure that the important frequencies were not transmitted to the capacitor, if you felt that it was a problem, and you could reevaluate the vibration in the rest of the board to see how it was affected by the extra coupled mass of the capacitor.
I guess the most important point is that whether something like this works well or not isn't really a material property, you have to take the whole system into account.
Rant over.... time for lunch!

02-05-07, 10:11 AM
I think the general view is that to many smr's can lead to a slightly sterile sound, thats what I found when I tried .Try using BC128's for coupling Les should have some, and I think feedback MMK's should really be 47uf.

02-05-07, 10:41 AM
With a 1K resistor in the feedback loop you do need 47uF, but you can get away with 22uF once the resistor has been swapped to 2K.

The corner frequency of the 1K/47uF pair is 3.4Hz and the corner frequency of the 2K/22uF pair is 3.6Hz - not a substantial difference :)


03-05-07, 01:47 AM
The main reason why the corner frequency is so low is to reduce the nonlinear effects the electrolytic capacitor - this is described in Douglas Self's book.

When using a film cap, the nonlinearity is essentially eliminated, so you can probably get away with a somewhat higher corner frequency - I suspect 22uF and 1K (7.2Hz corner) would sound absolutely fine; it will be down 1dB at 15Hz, and only a tiny fraction of a dB anywhere there are musical fundamentals (bottom note on a concert grand is 27Hz).